Senior Capstone repo for integrating Raspberry Pi and neural computing stick for real time image processing
- Anaconda
- Clone the repo.
- Run
conda env create -f env.yml
- Run
conda activate piai-env
- Run
piai.py
Run conda env export > env.yml
- Gather training images
- Annotate training images using labelImg to get Pascal Voc format (.xml)
- use xml_to_csv.py to convert .xml annotations to .csv format
- Make test.record and train.record for Tensorflow to use in training
- Make a object_detection.pbtxt with labels
- Set paths and hyperparameters in pipeline.config file
- Train model using Tensorflow (train.py)
- View training results in Tensorboard
- Create frozen model (export_inference_graph.py)
- Run model on set of images (object_detection_tutorial.ipynb)
- Make adjustments to pipeline.config, add training images, retrain
- Put frozen model on USB, connect to RPi, copy frozen model to RPi
- Use openVINO model optimizer to convert frozen_inference_graph.pb file to RPi compatible format
- Run model on RPi using Pi camera
- From drone, connect battery and use remote connection
- View results in real time
- model optimizer: https://docs.openvinotoolkit.org/latest/openvino_docs_MO_DG_prepare_model_convert_model_Convert_Model_From_TensorFlow.html
- install for RPi: https://docs.openvinotoolkit.org/latest/openvino_docs_install_guides_installing_openvino_raspbian.html
- model zoo: https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf1_detection_zoo.md
- install: https://docs.anaconda.com/anaconda/user-guide/tasks/tensorflow/
- install: https://pypi.org/project/labelImg/
- install guide: https://www.youtube.com/watch?v=RplXYjxgZbw&list=PL0AA7mM1J4JtT23k6HFYOCq0EpCFsQi4a&index=4
- version compatibility: https://www.tensorflow.org/install/source
Guide for TensorFlow 1: